PHP-Friends vServer L
So, I have been using this vServer L for a while (over a year now) from @PHP_Friends.
See the yabs below for pure numbery stuff however here are some things I wanted to say.
Networking and peering is great, this server is connected to 10 GBit/s even though it isn't stated on the website (1GBit/s with 20TB/m is what I ordered)!
CPU performance was also really really good, never had a problem with that. What I really liked about it is the enabled nested virtualization.
What surprised me tho, especially when yabsing it now, is the (fluctuating) disk performance. Apparently PHP-Friends has a fixed IO guarantee https://php-friends.de/vserver-ssd/io-garantie, so I think one can always rely to get these numbers, I did not reach them with yabs tho. They are guaranteed only with the use of their fio command. I have not experienced any IO problems with the server, so that was perfectly fine for me.
However the best thing, at last, is the support. It's fast, it's quick, it's problem solving, it's awesome. If you know other German providers (looking at you netcup), PHP-Friends was especially very helpful when it comes to contract changes, shout-out to that!
They increased the price from 25€ to now 28€ a month in January, however I think it's a decent deal as a general available offer.
Final verdict: Would recommend, stable host
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
# Yet-Another-Bench-Script #
# v2023-03-24 #
# https://github.com/masonr/yet-another-bench-script #
# ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
Mi 12. Apr 17:34:36 CEST 2023
Basic System Information:
---------------------------------
Uptime : 0 days, 0 hours, 6 minutes
Processor : AMD EPYC 7452 32-Core Processor
CPU cores : 8 @ 2349.998 MHz
AES-NI : ✔ Enabled
VM-x/AMD-V : ✔ Enabled
RAM : 39.3 GiB
Swap : 0.0 KiB
Disk : 250.1 GiB
Distro : Debian GNU/Linux 10 (buster)
Kernel : 4.19.0-5-amd64
VM Type : KVM
Net Online : IPv4 & IPv6
IPv6 Network Information:
---------------------------------
ISP : dataforest GmbH
ASN : AS58212 dataforest GmbH
Host : dataforest GmbH
Location : Frankfurt am Main, Hesse (HE)
Country : Germany
fio Disk Speed Tests (Mixed R/W 50/50):
---------------------------------
Block Size | 4k (IOPS) | 64k (IOPS)
------ | --- ---- | ---- ----
Read | 32.10 MB/s (8.0k) | 940.91 MB/s (14.7k)
Write | 32.15 MB/s (8.0k) | 945.86 MB/s (14.7k)
Total | 64.25 MB/s (16.0k) | 1.88 GB/s (29.4k)
| |
Block Size | 512k (IOPS) | 1m (IOPS)
------ | --- ---- | ---- ----
Read | 1.00 GB/s (1.9k) | 935.28 MB/s (913)
Write | 1.05 GB/s (2.0k) | 997.57 MB/s (974)
Total | 2.05 GB/s (4.0k) | 1.93 GB/s (1.8k)
iperf3 Network Speed Tests (IPv4):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 6.56 Gbits/sec | 8.51 Gbits/sec | 14.8 ms
Scaleway | Paris, FR (10G) | 6.87 Gbits/sec | 3.98 Gbits/sec | 12.1 ms
NovoServe | North Holland, NL (40G) | 8.05 Gbits/sec | 7.17 Gbits/sec | 6.58 ms
Uztelecom | Tashkent, UZ (10G) | 1.96 Gbits/sec | 1.65 Gbits/sec | 77.2 ms
Clouvider | NYC, NY, US (10G) | 2.15 Gbits/sec | 801 Mbits/sec | 78.6 ms
Clouvider | Dallas, TX, US (10G) | 1.49 Gbits/sec | 165 Mbits/sec | 124 ms
Clouvider | Los Angeles, CA, US (10G) | 1.25 Gbits/sec | 32.5 Mbits/sec | 153 ms
iperf3 Network Speed Tests (IPv6):
---------------------------------
Provider | Location (Link) | Send Speed | Recv Speed | Ping
----- | ----- | ---- | ---- | ----
Clouvider | London, UK (10G) | 7.65 Gbits/sec | 2.20 Gbits/sec | 11.7 ms
Scaleway | Paris, FR (10G) | busy | busy | 9.92 ms
NovoServe | North Holland, NL (40G) | busy | busy | 6.57 ms
Uztelecom | Tashkent, UZ (10G) | 2.34 Gbits/sec | busy | 73.8 ms
Clouvider | NYC, NY, US (10G) | 2.25 Gbits/sec | 1.70 Gbits/sec | 78.7 ms
Clouvider | Dallas, TX, US (10G) | 1.43 Gbits/sec | 1.51 Gbits/sec | 124 ms
Clouvider | Los Angeles, CA, US (10G) | 1.13 Gbits/sec | 255 Mbits/sec | 151 ms
Geekbench 6 Benchmark Test:
---------------------------------
Test | Value
|
Single Core | 1055
Multi Core | 4687
Full Test | https://browser.geekbench.com/v6/cpu/890411
Comments
Hi,
the 4k IOPS should be more, can you open a ticket about this? We can apply various tuning
Mine so much better though.
PHP-Friends 2 Cores AMD 10GB RAM - 9.99 EURO
Sun 09 Apr 2023 11:06:30 AM CEST Basic System Information: --------------------------------- Uptime : 0 days, 0 hours, 48 minutes Processor : AMD EPYC 7452 32-Core Processor CPU cores : 2 @ 2349.998 MHz AES-NI : ✔ Enabled VM-x/AMD-V : ✔ Enabled RAM : 9.7 GiB Swap : 1024.0 MiB Disk : 77.8 GiB Distro : Ubuntu 20.04.6 LTS Kernel : 5.4.0-146-generic VM Type : KVM Net Online : IPv4 & IPv6 IPv6 Network Information: --------------------------------- ISP : dataforest GmbH ASN : AS58212 dataforest GmbH Host : dataforest GmbH Location : Frankfurt am Main, Hesse (HE) Country : Germany fio Disk Speed Tests (Mixed R/W 50/50): --------------------------------- Block Size | 4k (IOPS) | 64k (IOPS) ------ | --- ---- | ---- ---- Read | 249.03 MB/s (62.2k) | 3.46 GB/s (54.1k) Write | 249.69 MB/s (62.4k) | 3.48 GB/s (54.4k) Total | 498.72 MB/s (124.6k) | 6.95 GB/s (108.6k) | | Block Size | 512k (IOPS) | 1m (IOPS) ------ | --- ---- | ---- ---- Read | 8.44 GB/s (16.4k) | 8.08 GB/s (7.8k) Write | 8.89 GB/s (17.3k) | 8.62 GB/s (8.4k) Total | 17.33 GB/s (33.8k) | 16.71 GB/s (16.3k) iperf3 Network Speed Tests (IPv4): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | Ping ----- | ----- | ---- | ---- | ---- Clouvider | London, UK (10G) | 5.95 Gbits/sec | 2.55 Gbits/sec | 11.5 ms Scaleway | Paris, FR (10G) | busy | busy | 12.5 ms NovoServe | North Holland, NL (40G) | 3.17 Gbits/sec | 3.57 Gbits/sec | 6.32 ms Uztelecom | Tashkent, UZ (10G) | 1.95 Gbits/sec | 1.91 Gbits/sec | 76.1 ms Clouvider | NYC, NY, US (10G) | 1.25 Gbits/sec | 2.95 Gbits/sec | 82.5 ms Clouvider | Dallas, TX, US (10G) | 1.40 Gbits/sec | 1.13 Gbits/sec | 120 ms Clouvider | Los Angeles, CA, US (10G) | 1.12 Gbits/sec | 1.95 Gbits/sec | 146 ms iperf3 Network Speed Tests (IPv6): --------------------------------- Provider | Location (Link) | Send Speed | Recv Speed | Ping ----- | ----- | ---- | ---- | ---- Clouvider | London, UK (10G) | 8.12 Gbits/sec | 5.60 Gbits/sec | 11.5 ms Scaleway | Paris, FR (10G) | 6.97 Gbits/sec | 8.42 Gbits/sec | 9.64 ms NovoServe | North Holland, NL (40G) | 8.04 Gbits/sec | 3.94 Gbits/sec | 6.25 ms Uztelecom | Tashkent, UZ (10G) | 2.63 Gbits/sec | 2.47 Gbits/sec | 76.1 ms Clouvider | NYC, NY, US (10G) | 2.28 Gbits/sec | 2.28 Gbits/sec | 78.9 ms Clouvider | Dallas, TX, US (10G) | 1.46 Gbits/sec | 1.49 Gbits/sec | 117 ms Clouvider | Los Angeles, CA, US (10G) | 999 Mbits/sec | 1.06 Gbits/sec | 147 ms YABS completed in 7 min 55 secI use the vServer S SSD G3 AMD plan priced 9.99 EURO a month.
For more detailed benchmark:
https://kangserver.id/benchmark-vps-php-friends-2-cores-amd-10gb-ram-harga-9-99-euro-2023
There is no doubt with them, they are very professional. If having problems they will contact you for workarounds or offer an option/solution. Which is mainly the best feature they have aside from offering best connectivity, good/real resources as advertised
does the PHP-Friends IO warranty also apply to avoro (https://php-friends.de/vserver-ssd/io-garantie)? If so, what is your professional opinion about this machine, especially concerning these values: 64K, 512K, 1m especially in comparison with the aforementioned values of @febryanvaldo? @PHP_Friends @Avoro
Quick reply, love that team.
Well, I think it's easily resolvable, however I have cancelled the server already so there is no need anymore. Thanks for the offer tho!
Those other yabses above look just like what I would have expected. Maybe it's due to the template I used (just reinstalled for the yabs
)
well... fio Disk Speed Test from my vps are noticeably weaker than from @febryanvaldo vps! sadly no quick reply from @Avoro and @PHP_Friends until now.
Your disk speeds are fine, even more than fine. Sure, they are not outrageous high amazing numbers dick contest measurements, but they are fuuuuucking fast for 99.99% of usages and because you are asking that kind of question here rather raising issue with support I can tell you don't even notice those in daily usage
Well, these are just normal SATA SSDs, this is not to be compared with NVMe SSDs
you may be right... i wanted to take advantage of the moment to point out the good but not very good values. febry's values, however, are much better.
sir... with respect
@PHP_Friends @Avoro
all avoro vServer are, according to the avoro website, equipped with nvme memory. is the own portfolio known?
Yes, I am aware of our products, but it doesn't change the fact that you have normal SSDs since this is our old Compute vServer line. The new one is NVMe only and runs on our hyperconverged Ceph.
...or do avoro special deals have "only" ssds? i can also live with this. i only had the faint hope that i might also be able to benefit from nvme memory.
However, to be clear, all this does not change the fact that dataforest is absolutely premium.
It has nothing to do with any deals, you are talking about a SSD product that has been offered with SSDs till we moved to our NVMe Ceph with it.
Can we just agree on the fact that 151k random IOPS are perfectly fine, no matter if it's SSD or NVMe?
Of course the sequential speeds are "worse" than with NVMe. And to be clear, the IOPS guarantee only applies to the PHP-Friends products but will be reached with the Avoro products too usually.
sir, i think i need to go into more detail about my product. it's a compute vserver (https://avoro.eu/de/compute-vserver/) that was set up pretty much 6 months ago. Question: were the compute vserver six months ago still equipped with ssds and not with nvme memory?
of course we can agree on that. i am not interested in any escalation!
https://web.archive.org/web/20221202151839/https://avoro.eu/en/compute-vserver/
xx GB SAS-SSD 👍🏻
I'd like to try it out if the bgp sessions fee can be waived.
Unfortunately not, sorry
@PHP_Friends
Do you accept Asian customers please, in the last week NETCUP suspended the servers of all Chinese customers, estimated to be in the thousands, without notice. The funny thing is that they continued these customers to invoice for the next month's bill after doing this. I wonder if you, as a German company, would do the same thing to your customers?
I had asked them through website bt never got replied.
Hi,
we have a lot of asian customers, that's not a problem.
I hardly think so, we answer every customer. Maybe you entered a wrong mail?
Will do it once again. I thought its because of language.